Overclock.net › Forums › Industry News › Hardware News › [YT] Frame Rating and FCAT Explanation and Discussion with NVIDIA
New Posts  All Forums:Forum Nav:

[YT] Frame Rating and FCAT Explanation and Discussion with NVIDIA - Page 14  

post #131 of 311
Quote:
Originally Posted by thestache View Post

Do you know how to have a discussion? Because I'm not sure you do. All you're doing is attempting to tell me something that has no relevance to what I'm talking about and is just recycled garbage from that video.

How do you know drivers don't effect it? You done tests to prove it? How do you know every setup suffers from the exact same percentage of frame timing variances? How do you know every un metered setup has it at all times? How many un-metered multi GPU setups have you had in the last 2-3 years? I've had six different ones in the last few years since the HD 6000 series and not all of them had perceivable microstutter or any of the rubbish you're referring to. Not to mention your signature rig only has a single GPU. So where does you experience with the HD 7000 and HD 6000 series come from? Because I've owned the HD 6990, HD 6970, HD 7990 Devil 13 and HD 7970 all in 3-4 Way GPU setups across many different driver revisions and not all of them had perceivable microstutter that was an issue and I can spot micro stutter with my single GTX Titan.

Not to mention you're completly wrong about if one GPU can pull off 60 FPS you won't notice it.

Ok, I didn't succeed in explaining it to you and I dun think I can explain any better without posting a bunch of graphs which I won't do. You have everything in the latest PCper reviews.

And no, I don't own any CF setups, if I did I would test it (altho I would have to use only my eyes to measure). But numbers don't lie. 'Observed FPS' graphs are there to be seen. Also I hope u understand how dropped and runt frames appear, why they exist, b/c I think that is one of the basics here to understand the rest.

And on what do you base your belief that some previous AMD driver would change anything if there was never frame-metering? Explain at least that.
Edited by Ha-Nocri - 5/11/13 at 2:10am
post #132 of 311
Quote:
Originally Posted by Ha-Nocri View Post

It's not that complicated at all. Let me try to explain.

There was no frame metering with AMD EVER. So using previous drivers changes nothing.

What ppl thought was micro-stuttering was just under-performing. They would see 60 fps using FRAPS or other program, but in reality they were on 35 fps or so, equal to what 1 GPU could do. And we all know 35 fps feels very 'laggy' compared to 60 fps.

There are times you don't notice it. Why? B/c those games are not GPU-intensive so 1 GPU can pull 50-60 fps off. So even tho FRAPS would show 100 fps (or 60 fps v-sync on) you were getting 60 fps and things looked smooth. That is why some games are indeed smooth with CF. But those games like Crysis that need much GPU power will be very noticeable b/c you are actually playing @ 20-30 fps which is not enough.

To sum it up, if 1 GPU can pull off 50-60 fps, you will see no CF issues.

So, I am not taking frame-time variance into account 'cause I think avg. fps is much more important. Sure, it is frame-variance that causes dropped and runt frames, but above that only 'observed' fps is what really matters.

I disagree. Observed fps would be The Metric if there was an universally agreed standard on what qualifies as an observed frame. There isn't. It's a judgement call by PCPer (or nVidia, not sure) which states that any frame that shows as 20 scan lines or less on a 60Hz display is considered a runt (or dropped if 0). As of yet, there is no easy answers to the question of what makes smooth real-time rendering. It's a combination of frame rate and timely distribution. The complexity of all this is a huge disadvantage for frame rating compared to the one-number-to-rule-them-all fps method. It's a long journey ahead before it gets similar hold on the minds of people.
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
The Air Tunnel
(10 items)
 
  
CPUMotherboardGraphicsRAM
i7-4820K (4500MHz@1.28V) P9X79 Sapphire R9 290 Tri-X New Edition Team Group Vulcan 4x4GB 2133MHz 
Hard DriveCoolingOSPower
2x240GB SSD @RAID 0 Noctua NH-D14 SE2011 Windows 10 Pro Corsair AX750 
CaseAudio
Rosewill Armor EVO hiFace+AudioGD+DT770Pro 
  hide details  
post #133 of 311
Quote:
Originally Posted by specopsFI View Post

I disagree. Observed fps would be The Metric if there was an universally agreed standard on what qualifies as an observed frame. There isn't. It's a judgement call by PCPer (or nVidia, not sure) which states that any frame that shows as 20 scan lines or less on a 60Hz display is considered a runt (or dropped if 0). As of yet, there is no easy answers to the question of what makes smooth real-time rendering. It's a combination of frame rate and timely distribution. The complexity of all this is a huge disadvantage for frame rating compared to the one-number-to-rule-them-all fps method. It's a long journey ahead before it gets similar hold on the minds of people.

Agree... at least to some extent. Yes, there is that limit of 20 lines that had to be made, they have all the testing equipment needed so I tend to believe their judgement.

There is no easy answer indeed, I agree, but 'observed fps' is what matters the most, the nVidia guy said the same. I don't just repeat what he said w/o understanding (what some ppl here obviously believe), I just think the same. He said that almost universally higher real FPS is better. You can watch a video v-sync ON vs. v-sync OFF of a game-play @ ~50 fps. Even tho with v-sync ON you have terrible frame-time variance, smoothness doesn't change much between the 2.
post #134 of 311
Quote:
Originally Posted by Ha-Nocri View Post

Ok, I didn't succeed in explaining it to you and I dun think I can explain any better without posting a bunch of graphs which I won't do. You have everything in the latest PCper reviews.

And no, I don't own any CF setups, if I did I would test it (altho I would have to use only my eyes to measure). But numbers don't lie. 'Observed FPS' graphs are there to be seen. Also I hope u understand how dropped and runt frames appear, why they exist, b/c I think that is one of the basics here to understand the rest.

And on what do you base your belief that some previous AMD driver would change anything if there was never frame-metering? Explain at least that.

All you're doing is recycling information about observed FPS and how you seem to think multi GPU setups work or should I say don't work in this case. I understand the term observed FPS, did before PCPerspective started all this and don't care to discuss the rest because it has no relevance to what I was originally talking about, not to mention your view of it and AMD cards is incredibly ignorant. Confirmed by yourself with your zero experience.

The fact I question if different driver revisions effect what PCPerspective is reporting is because I've used different driver revisions in the past and had them effect micro stutter in positive and negative ways before for both Nvidia and AMD with un metered cards. As I'm sure many others have and I would like to see perceivable and non perceivable differences backed up by capture card and FCAT frame metering data.

And yes you do just repost what you think you heard. Because you don't understand how un metered multi GPU setups and runts work. By your understanding with crossfire every second frame is a runt but as long as the first GPU can handle 60FPS on its own you won't notice the runts. Which is wrong.

Observed FPS and keeping it above 60FPS isn't important. It just helps us understand to what extent the microstutter is effecting performance. Frame time variance is the important thing here. Regardless of your FPS if you frame time variance is consistent and always within 1-2ms of deviation from the frame timing relative to the on screen FPS it doesn't matter what your FPS is because you won't be loosing performance to microstutter and you will feel the FPS your GPU is outputting.

Re watch the video because you're focusing on the wrong things. The graph 'frame latincies per percentile' is what you need to focus on. You want that graph to be as flat as possible for as long as possible because that will ammount to an observed FPS equal to on screen FPS.
Edited by thestache - 5/11/13 at 3:57am
    
CPUMotherboardGraphicsRAM
i7 4770k Gigabyte Z87X OC  Gigabyte GTX Titan Gskill Ripjaws 2400mhz 16GB 4x4GB  
Hard DriveHard DriveCoolingCooling
Corsair Force GS 240GB Western Digital Velociraptor 320GB EK GTX Titan Acetal Copper FC Block with Backplate EK Monarch DIMM Module heatspreaders 
CoolingCoolingCoolingOS
EK X3 Reservoir 150 EK Gigabyte Z87 X OC Acetal Copper Block EK Monarch X4 Copper Acetal Block with clean top Windows 8 64-Bit 
MonitorMonitorMonitorKeyboard
Dell U2412M Dell U2412M Dell U2412M Razer BlackWidow 
PowerCaseMouseMouse Pad
Corsair AX1200i Custome Acetal case CNC Machined from 8mm sheet  Razer Mamba Razer Ironclad 
OtherOtherOtherOther
Corsair Individually Sleeved Cables Koolance 90deg Elbows  Monsoon Compression Fittings Corsair SP120 Quiet Edition PWM Fans 
OtherOtherOtherOther
Koolance OD3 Quick Disconnect Male Compression Koolance OD3 Quick Disconnect Female G1/4 Threaded EK Acetal Clean Supremacy Block XSPC RX360 Radiator 
OtherOtherOther
XSPC Acrylic Tank Reservior/Pump Top XSPC RX360 Radiator  Koolance 450S D5 Pump 
  hide details  
    
CPUMotherboardGraphicsRAM
i7 4770k Gigabyte Z87X OC  Gigabyte GTX Titan Gskill Ripjaws 2400mhz 16GB 4x4GB  
Hard DriveHard DriveCoolingCooling
Corsair Force GS 240GB Western Digital Velociraptor 320GB EK GTX Titan Acetal Copper FC Block with Backplate EK Monarch DIMM Module heatspreaders 
CoolingCoolingCoolingOS
EK X3 Reservoir 150 EK Gigabyte Z87 X OC Acetal Copper Block EK Monarch X4 Copper Acetal Block with clean top Windows 8 64-Bit 
MonitorMonitorMonitorKeyboard
Dell U2412M Dell U2412M Dell U2412M Razer BlackWidow 
PowerCaseMouseMouse Pad
Corsair AX1200i Custome Acetal case CNC Machined from 8mm sheet  Razer Mamba Razer Ironclad 
OtherOtherOtherOther
Corsair Individually Sleeved Cables Koolance 90deg Elbows  Monsoon Compression Fittings Corsair SP120 Quiet Edition PWM Fans 
OtherOtherOtherOther
Koolance OD3 Quick Disconnect Male Compression Koolance OD3 Quick Disconnect Female G1/4 Threaded EK Acetal Clean Supremacy Block XSPC RX360 Radiator 
OtherOtherOther
XSPC Acrylic Tank Reservior/Pump Top XSPC RX360 Radiator  Koolance 450S D5 Pump 
  hide details  
post #135 of 311
Quote:
Originally Posted by thestache View Post

All you're doing is recycling information about observed FPS and how you seem to think multi GPU setups work or should I say don't work in this case. I understand the term observed FPS, did before PCPerspective started all this and don't care to discuss the rest because it has no relevance to what I was originally talking about, not to mention your view of it and AMD cards is incredibly ignorant. Confirmed by yourself with your zero experience.

The fact I question if different driver revisions effect what PCPerspective is reporting is because I've used different driver revisions in the past and had them effect micro stutter in positive and negative ways before for both Nvidia and AMD with un metered cards. As I'm sure many others have and I would like to see perceivable and non perceivable differences backed up by capture card and FCAT frame metering data.

And yes you do just repost what you think you heard. Because you don't understand how un metered multi GPU setups and runts work. By your understanding with crossfire every second frame is a runt but as long as the first GPU can handle 60FPS on its own you won't notice the runts. Which is wrong.

Observed FPS and keeping it above 60FPS isn't important. It just helps us understand to what extent the microstutter is effecting performance. Frame time variance is the important thing here. Regardless of your FPS if you frame time variance is consistent and always within 1-2ms of deviation from the frame timing relative to the on screen FPS it doesn't matter what your FPS is because you won't be loosing performance to microstutter and you will feel the FPS your GPU is outputting.

Re watch the video because you're focusing on the wrong things. The graph 'frame latincies per percentile' is what you need to focus on. You want that graph to be as flat as possible for as long as possible because that will ammount to an observed FPS equal to on screen FPS.

Dude, you made me spilled my coffee! lachen.gif
    
CPUMotherboardGraphicsGraphics
i7 920 Evga X58 Classified LE Edition. Sapphire HD 7950 OC Edition Sapphire HD 7950 OC Edition 
RAMHard DriveHard DriveOptical Drive
G skill Ripjaws 12 gb Samsung ssd 120 gb sata III WD Velociraptop 300 gb x 2 LG Blu ray drive 
CoolingCoolingCoolingCooling
EK-FC7950 with EK-7950 Backplate FC8 30 watt per channel Bitspower fittings Feser 360 rad (Push Pull) 
CoolingCoolingCoolingCooling
XSPC 240 rad (Push Pull) Corsair SP 120 mm static x10 Corsair AF 120 mm  Bitspower 250 reservoir 
CoolingCoolingCoolingCooling
Heat killer 3.0 Koolance chipset water block EK-FC7950 with EK-7950 Backplate MCP35X2 with stand 
OSMonitorKeyboardPower
Windows 7 Ultimate Acer 27" x 3 (eyefinity) Logitech G510/CM Storm Mechanical Keyboard Seasonic 1250 watt X series with moddyi Cable set. 
CaseMouseMouse PadAudio
CaseLabs SM8 white ZowieAM Rocketfish Gamecon 780 
Audio
Bose 2.1 
  hide details  
    
CPUMotherboardGraphicsGraphics
i7 920 Evga X58 Classified LE Edition. Sapphire HD 7950 OC Edition Sapphire HD 7950 OC Edition 
RAMHard DriveHard DriveOptical Drive
G skill Ripjaws 12 gb Samsung ssd 120 gb sata III WD Velociraptop 300 gb x 2 LG Blu ray drive 
CoolingCoolingCoolingCooling
EK-FC7950 with EK-7950 Backplate FC8 30 watt per channel Bitspower fittings Feser 360 rad (Push Pull) 
CoolingCoolingCoolingCooling
XSPC 240 rad (Push Pull) Corsair SP 120 mm static x10 Corsair AF 120 mm  Bitspower 250 reservoir 
CoolingCoolingCoolingCooling
Heat killer 3.0 Koolance chipset water block EK-FC7950 with EK-7950 Backplate MCP35X2 with stand 
OSMonitorKeyboardPower
Windows 7 Ultimate Acer 27" x 3 (eyefinity) Logitech G510/CM Storm Mechanical Keyboard Seasonic 1250 watt X series with moddyi Cable set. 
CaseMouseMouse PadAudio
CaseLabs SM8 white ZowieAM Rocketfish Gamecon 780 
Audio
Bose 2.1 
  hide details  
post #136 of 311
Quote:
Originally Posted by thestache View Post

All you're doing is recycling information about observed FPS and how you seem to think multi GPU setups work or should I say don't work in this case. I understand the term observed FPS, did before PCPerspective started all this and don't care to discuss the rest because it has no relevance to what I was originally talking about, not to mention your view of it and AMD cards is incredibly ignorant. Confirmed by yourself with your zero experience.

The fact I question if different driver revisions effect what PCPerspective is reporting is because I've used different driver revisions in the past and had them effect micro stutter in positive and negative ways before for both Nvidia and AMD with un metered cards. As I'm sure many others have and I would like to see perceivable and non perceivable differences backed up by capture card and FCAT frame metering data.

And yes you do just repost what you think you heard. Because you don't understand how un metered multi GPU setups and runts work. By your understanding with crossfire every second frame is a runt but as long as the first GPU can handle 60FPS on its own you won't notice the runts. Which is wrong.

Observed FPS and keeping it above 60FPS isn't important. It just helps us understand to what extent the microstutter is effecting performance. Frame time variance is the important thing here. Regardless of your FPS if you frame time variance is consistent and always within 1-2ms of deviation from the frame timing relative to the on screen FPS it doesn't matter what your FPS is because you won't be loosing performance to microstutter and you will feel the FPS your GPU is outputting.

Re watch the video because you're focusing on the wrong things. The graph 'frame latincies per percentile' is what you need to focus on. You want that graph to be as flat as possible for as long as possible because that will ammount to an observed FPS equal to on screen FPS.

Ok, you obviously educated yourself between my last response to you and this post of yours. Good. I hope it won't take you much to agree with me... if you'll be willing to admit it ofc. Also, no, not every 2nd frame is a runt, it's more complicated. Not all frame-times are the same.

Cheers thumb.gif
post #137 of 311
Quote:
Originally Posted by Ha-Nocri View Post

Ok, you obviously educated yourself between my last response to you and this post of yours. Good. I hope it won't take you much to agree with me... if you'll be willing to admit it ofc. Also, no, not every 2nd frame is a runt, it's more complicated. Not all frame-times are the same.

Cheers thumb.gif

Sorry but to me, you make no sense. You assume too much, then bring up things that are irrelevant, wonder off, are vague then not specific enough and your chain of thought is just too irratic for me to keep up with. I didn't brush up on any reading what so ever, maybe you just finally understood something I was saying. Regardless this is going no where, so you have a nice day.
    
CPUMotherboardGraphicsRAM
i7 4770k Gigabyte Z87X OC  Gigabyte GTX Titan Gskill Ripjaws 2400mhz 16GB 4x4GB  
Hard DriveHard DriveCoolingCooling
Corsair Force GS 240GB Western Digital Velociraptor 320GB EK GTX Titan Acetal Copper FC Block with Backplate EK Monarch DIMM Module heatspreaders 
CoolingCoolingCoolingOS
EK X3 Reservoir 150 EK Gigabyte Z87 X OC Acetal Copper Block EK Monarch X4 Copper Acetal Block with clean top Windows 8 64-Bit 
MonitorMonitorMonitorKeyboard
Dell U2412M Dell U2412M Dell U2412M Razer BlackWidow 
PowerCaseMouseMouse Pad
Corsair AX1200i Custome Acetal case CNC Machined from 8mm sheet  Razer Mamba Razer Ironclad 
OtherOtherOtherOther
Corsair Individually Sleeved Cables Koolance 90deg Elbows  Monsoon Compression Fittings Corsair SP120 Quiet Edition PWM Fans 
OtherOtherOtherOther
Koolance OD3 Quick Disconnect Male Compression Koolance OD3 Quick Disconnect Female G1/4 Threaded EK Acetal Clean Supremacy Block XSPC RX360 Radiator 
OtherOtherOther
XSPC Acrylic Tank Reservior/Pump Top XSPC RX360 Radiator  Koolance 450S D5 Pump 
  hide details  
    
CPUMotherboardGraphicsRAM
i7 4770k Gigabyte Z87X OC  Gigabyte GTX Titan Gskill Ripjaws 2400mhz 16GB 4x4GB  
Hard DriveHard DriveCoolingCooling
Corsair Force GS 240GB Western Digital Velociraptor 320GB EK GTX Titan Acetal Copper FC Block with Backplate EK Monarch DIMM Module heatspreaders 
CoolingCoolingCoolingOS
EK X3 Reservoir 150 EK Gigabyte Z87 X OC Acetal Copper Block EK Monarch X4 Copper Acetal Block with clean top Windows 8 64-Bit 
MonitorMonitorMonitorKeyboard
Dell U2412M Dell U2412M Dell U2412M Razer BlackWidow 
PowerCaseMouseMouse Pad
Corsair AX1200i Custome Acetal case CNC Machined from 8mm sheet  Razer Mamba Razer Ironclad 
OtherOtherOtherOther
Corsair Individually Sleeved Cables Koolance 90deg Elbows  Monsoon Compression Fittings Corsair SP120 Quiet Edition PWM Fans 
OtherOtherOtherOther
Koolance OD3 Quick Disconnect Male Compression Koolance OD3 Quick Disconnect Female G1/4 Threaded EK Acetal Clean Supremacy Block XSPC RX360 Radiator 
OtherOtherOther
XSPC Acrylic Tank Reservior/Pump Top XSPC RX360 Radiator  Koolance 450S D5 Pump 
  hide details  
post #138 of 311
What's the deal with alot of AMD users?

Denying because they truly are ignorant of the issue?

Or

Denying because they don't want to admit to themselves that they bought a broken product?

I really don't see why there's so much back and forth about this. The CF problem exists. It's been confirmed by many different people on many different setups. It's real. It's inherent to AMD tech. If you own AMD, then there's a 100% you have this issue. But really, I don't see why AMD owners are going to just straight up deny that there's some kind of issue? I'd be pissed and trying to knock AMD's door down for answers and a fix... Not trying to bury my head in the sand.
post #139 of 311
Quote:
Originally Posted by lajgnd View Post

What's the deal with alot of AMD users?

Denying because they truly are ignorant of the issue?

Or

Denying because they don't want to admit to themselves that they bought a broken product?

I really don't see why there's so much back and forth about this. The CF problem exists. It's been confirmed by many different people on many different setups. It's real. It's inherent to AMD tech. If you own AMD, then there's a 100% you have this issue. But really, I don't see why AMD owners are going to just straight up deny that there's some kind of issue? I'd be pissed and trying to knock AMD's door down for answers and a fix... Not trying to bury my head in the sand.

Denying? My crossfire 7950 is on air at first, they were so smooth at any games I throw at them and decided to put them on water to squeeze more juice out of them because I play on 5670x1080. The smoothness is just amazing to me. I also owned a gtx 690 but that one is on htpc that my daughter use for league of legends. My experience on 7950 is more smooth than 690 with eyefinity setup. Who knows maybe my crossfire are NVidia cars in disguise......I did some investigating last night with a magnifying glass on the pcb. I could not find anything that they are NVidia gpu cards in disguise.....gunner.gif
    
CPUMotherboardGraphicsGraphics
i7 920 Evga X58 Classified LE Edition. Sapphire HD 7950 OC Edition Sapphire HD 7950 OC Edition 
RAMHard DriveHard DriveOptical Drive
G skill Ripjaws 12 gb Samsung ssd 120 gb sata III WD Velociraptop 300 gb x 2 LG Blu ray drive 
CoolingCoolingCoolingCooling
EK-FC7950 with EK-7950 Backplate FC8 30 watt per channel Bitspower fittings Feser 360 rad (Push Pull) 
CoolingCoolingCoolingCooling
XSPC 240 rad (Push Pull) Corsair SP 120 mm static x10 Corsair AF 120 mm  Bitspower 250 reservoir 
CoolingCoolingCoolingCooling
Heat killer 3.0 Koolance chipset water block EK-FC7950 with EK-7950 Backplate MCP35X2 with stand 
OSMonitorKeyboardPower
Windows 7 Ultimate Acer 27" x 3 (eyefinity) Logitech G510/CM Storm Mechanical Keyboard Seasonic 1250 watt X series with moddyi Cable set. 
CaseMouseMouse PadAudio
CaseLabs SM8 white ZowieAM Rocketfish Gamecon 780 
Audio
Bose 2.1 
  hide details  
    
CPUMotherboardGraphicsGraphics
i7 920 Evga X58 Classified LE Edition. Sapphire HD 7950 OC Edition Sapphire HD 7950 OC Edition 
RAMHard DriveHard DriveOptical Drive
G skill Ripjaws 12 gb Samsung ssd 120 gb sata III WD Velociraptop 300 gb x 2 LG Blu ray drive 
CoolingCoolingCoolingCooling
EK-FC7950 with EK-7950 Backplate FC8 30 watt per channel Bitspower fittings Feser 360 rad (Push Pull) 
CoolingCoolingCoolingCooling
XSPC 240 rad (Push Pull) Corsair SP 120 mm static x10 Corsair AF 120 mm  Bitspower 250 reservoir 
CoolingCoolingCoolingCooling
Heat killer 3.0 Koolance chipset water block EK-FC7950 with EK-7950 Backplate MCP35X2 with stand 
OSMonitorKeyboardPower
Windows 7 Ultimate Acer 27" x 3 (eyefinity) Logitech G510/CM Storm Mechanical Keyboard Seasonic 1250 watt X series with moddyi Cable set. 
CaseMouseMouse PadAudio
CaseLabs SM8 white ZowieAM Rocketfish Gamecon 780 
Audio
Bose 2.1 
  hide details  
post #140 of 311
Quote:
Originally Posted by Majin SSJ Eric View Post

Nah, Nvidia didn't have to pay PCPer to do these stories. Ryan hates AMD so much that he did them for free! tongue.gif

bahaha!
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
This thread is locked  
Overclock.net › Forums › Industry News › Hardware News › [YT] Frame Rating and FCAT Explanation and Discussion with NVIDIA